search

IDEA-Research / DINO

[ICLR 2023] Official implementation of the paper "DINO: DETR with Improved DeNoising Anchor Boxes for End-to-End Object Detection"
Apache License 2.0
2.19k stars 243 forks source link

num_classes #249

Open yuzhixiaobo opened 7 months ago

yuzhixiaobo commented 7 months ago

I see that the calculation of loss_cardinality includes a background class, so should num_class be set to the actual number of categories plus 1.

iiiidyllic commented 4 months ago

I see that the calculation of loss_cardinality includes a background class, so should num_class be set to the actual number of categories plus 1.

Have you tried it later? Does this make sense?

BikiniBottom1 commented 4 hours ago

我在目标检测的时候设置numclass为2（背景和目标），结果证明是对的